U.S. patent application number 12/198580 was filed with the patent office on 2010-03-04 for sensor for detecting and differentiating chemical analytes.
Invention is credited to Lawrence R. Senesac, Thomas G. Thundat, Dechang Yi.
Application Number | 20100055801 12/198580 |
Document ID | / |
Family ID | 41351509 |
Filed Date | 2010-03-04 |
United States Patent
Application |
20100055801 |
Kind Code |
A1 |
Yi; Dechang ; et
al. |
March 4, 2010 |
SENSOR FOR DETECTING AND DIFFERENTIATING CHEMICAL ANALYTES
Abstract
A sensor for detecting and differentiating chemical analytes
includes a microscale body having a first end and a second end and
a surface between the ends for adsorbing a chemical analyte. The
surface includes at least one conductive heating track for heating
the chemical analyte and also a conductive response track, which is
electrically isolated from the heating track, for producing a
thermal response signal from the chemical analyte. The heating
track is electrically connected with a voltage source and the
response track is electrically connected with a signal recorder.
The microscale body is restrained at the first end and the second
end and is substantially isolated from its surroundings
therebetween, thus having a bridge configuration.
Inventors: |
Yi; Dechang; (Metuchen,
NJ) ; Senesac; Lawrence R.; (Knoxville, TN) ;
Thundat; Thomas G.; (Knoxville, TN) |
Correspondence
Address: |
UT-Battelle/Chicago/BHGL
P.O. Box 10395
Chicago
IL
60610
US
|
Family ID: |
41351509 |
Appl. No.: |
12/198580 |
Filed: |
August 26, 2008 |
Current U.S.
Class: |
436/149 ;
422/90 |
Current CPC
Class: |
G01N 27/18 20130101 |
Class at
Publication: |
436/149 ;
422/90 |
International
Class: |
G01N 27/04 20060101
G01N027/04; G01N 33/22 20060101 G01N033/22; G01N 25/20 20060101
G01N025/20 |
Goverment Interests
FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT
[0001] This invention was made with government support under
Contract. No. DE-AC05-000R22725 awarded by the U.S. Department of
Energy. The government has certain rights in this invention.
Claims
1. A sensor for detecting and differentiating chemical analytes,
the sensor comprising: a microscale body having a first end and a
second end and a surface therebetween for adsorbing a chemical
analyte, the surface including at least one conductive heating
track for heating the chemical analyte and a conductive response
track electrically isolated from the heating track for producing a
thermal response signal from the chemical analyte, the heating
track being electrically connected with a voltage source and the
response track being electrically connected with a signal recorder,
wherein the microscale body is restrained at the first end and the
second end and is substantially isolated from its surroundings
therebetween, the sensor thereby having a bridge configuration.
2. The sensor of claim 1 wherein the microscale body has a
thickness of from about 50 nm to about 500 nm.
3. The sensor of claim 1 wherein the surface for adsorbing the
chemical analyte spans an area of from about 0.01 mm.sup.2 to about
0.1 mm.sup.2.
4. The sensor of claim 1 wherein each of the heating track and the
response track extends from the first end to the second end of the
microscale body.
5. The sensor of claim 4 wherein the microscale body includes two
conductive heating tracks disposed on either side of the conductive
response track.
6. The sensor of claim 1 wherein the conductive heating and
response tracks are embedded in the surface.
7. The sensor of claim 1 wherein the conductive heating and
response tracks are electrically isolated from each other by
silicon nitride.
8. The sensor of claim 1 wherein the conductive heating and
response tracks comprise doped silicon.
9. The sensor of claim 1 wherein the response track is electrically
connected to a Wheatstone bridge circuit including as a variable
resistor an unloaded microscale body free of adsorbed chemical
analyte, the Wheatstone bridge circuit providing a differential
output signal to the signal recorder.
10. The sensor of claim 1 wherein the chemical analyte is an
explosive.
11. The sensor of claim 10 wherein the explosive is selected from
the group consisting of TNT, PETN, and RDX.
12. A method of detecting and differentiating chemical analytes,
the method comprising: providing a microscale body having a first
end and a second end and a surface therebetween, the surface
including at least one conductive heating track and a conductive
response track electrically isolated from the heating track,
wherein the microscale body is restrained at the first end and the
second end and is substantially isolated from its surroundings
therebetween, the sensor thereby having a bridge configuration;
adsorbing a chemical analyte onto the surface of the microscale
body; applying an increasing voltage to the heating track to heat
the microscale body at a rate dT/dt for a time duration sufficient
to remove substantially all of the chemical analyte from the
surface; measuring a resistance of the response track during the
application of the increasing voltage over at least a portion of
the time duration of the heating to generate a plurality of thermal
response signals; and recording the plurality of thermal response
signals to obtain a first thermal response profile.
13. The method of claim 12 wherein the rate dT/dt is about
10.sup.4.degree. C./s.
14. The method of claim 12 wherein the chemical analyte is heated
to a temperature of at least about 500.degree. C. over the time
duration of the heating.
15. The method of claim 12 wherein the time duration of the heating
is from about 10 ms to about 100 ms.
16. The method of claim 12 wherein from about 100 picograms to
about 600 picograms of the chemical analyte are adsorbed onto the
surface of the microscale body.
17. The method of claim 12 further comprising collecting and
concentrating the chemical analyte prior to the adsorption using a
preconcentrator.
18. The method of claim 12 wherein applying the increasing voltage
comprises linearly ramping the voltage.
19. The method of claim 12 wherein measuring a resistance of the
response track comprises employing a Wheatstone bridge circuit to
make a differential measurement of the thermal response signal.
20. The method of claim 12 wherein the increasing voltage is
applied to the heating track a second time after removing
substantially all of the chemical analyte from the surface to
obtain a second thermal response profile, the second thermal
response profile being subtracted from the first thermal response
profile to obtain a thermal response profile of the chemical
analyte.
Description
TECHNICAL FIELD
[0002] The present disclosure is related generally to sensing
devices and more particularly to microscale sensors for detecting
chemical analytes.
BACKGROUND
[0003] Despite its immediate relevance in homeland security
applications, high sensitivity detection of explosives using
real-time, miniature sensors still remains as a crucial challenge.
Although widely used and highly effective, trace explosive
detection based on canines is neither cost effective nor suitable
for mass deployment. Currently available technologies, such as ion
mobility spectrometry and nuclear quadrupole resonance
spectroscopy, are bulky and expensive. Optical spectroscopic
techniques, such as Raman and laser-induced breakdown
spectroscopies, are highly selective but suffer from poor
sensitivity. Micro-electro-mechanical systems (MEMS) can
potentially satisfy many of the requirements for an ideal compact
chemical sensor, such as low-power consumption, real-time
operation, and high sensitivity. However, the suitability of MEMS
as practical sensors for vapor detection has traditionally been
limited by a lack of chemical selectivity.
[0004] The selectivity challenge encountered with micromechanical
sensors is not unique to MEMS. Other gravimetric sensors such as
quartz crystal microbalance (QCM) and surface acoustic wave (SAW)
devices, also lack intrinsic selectivity and rely on selective
interfaces for chemical speciation. The need for chemical
selectivity forces the use of separation techniques or the use of
highly selective recognition layers that are irreversible at room
temperature. Although attractive from an analytical standpoint,
incorporation of separation techniques with MEMS sensors poses
integration problems, especially for explosives detection due to
the large volumes of air needed for sample collection.
[0005] At present there exist no room-temperature reversible
receptors that are highly selective for vapor molecules, especially
explosive vapors. Designing high specificity molecular recognition
layers for small molecules is challenging due to the limited number
of chemical interactions that can serve as a basis for designing
selective layers while satisfying the highly desirable sensor
attribute of room temperature reversibility.
[0006] Approaches for achieving selectivity by using sensor arrays
modified with partially selective interfaces and pattern
recognition work are presently underway. The molecular recognition
interfaces based on weak interactions are not specific enough to
produce unique responses with a single sensor. Unique responses
(orthogonal) cannot be obtained if the mechanism behind individual
sensor elements is unspecific, for example, hydrogen bonding of
analyte with the chemoselective layer. Increasing the number of
sensor elements in the array for pattern recognition analysis can
improve the selectivity only if the responses from individual
sensing elements are orthogonal. However, there are only a limited
number of weak reversible chemical interactions that can serve as a
basis for designing the selective layers. Therefore, despite the
chemical sensing advantages offered by microfabricated sensors,
their use as a practical sensor may be limited without the
development of techniques that can generate orthogonal
responses.
BRIEF SUMMARY
[0007] A sensor and method that can provide unique thermal response
data for the detection and differentiation of explosives and other
chemical analytes are described.
[0008] The sensor includes a microscale body having a first end and
a second end and a surface between the ends for adsorbing a
chemical analyte. The surface includes at least one conductive
heating track for heating the chemical analyte and also a
conductive response track, which is electrically isolated from the
heating track, for producing a thermal response signal from the
chemical analyte. The heating track is electrically connected with
a voltage source and the response track is electrically connected
with a signal recorder. The microscale body is restrained at the
first end and the second end and is substantially isolated from its
surroundings therebetween, thus having a bridge configuration.
[0009] The method includes providing a microscale body having a
first end and a second end and a surface between the ends, where
the surface includes at least one conductive heating track and a
conductive response track, which is electrically isolated from the
heating track. The microscale body is restrained at the first end
and the second end and is substantially isolated from its
surroundings between the ends, thus having a bridge configuration.
The method includes adsorbing a chemical analyte onto the surface
of the microscale body and applying an increasing voltage to the
heating track to heat the microscale body for a time duration
sufficient to remove substantially all of the chemical analyte from
the surface. A resistance of the response track is measured during
the application of the increasing voltage over at least a portion
of the time duration of the heating to generate a plurality of
thermal response signals. The thermal response signals are recorded
to obtain a first thermal response profile.
BRIEF DESCRIPTION OF THE DRAWINGS
[0010] FIG. 1A is a schematic of one embodiment of the microbridge
sensor;
[0011] FIG. 1B shows the microbridge sensor of FIG. 1A as part of a
Wheatstone bridge circuit;
[0012] FIG. 2 is a scanning electron microscopy (SEM) image of the
sensor shown in FIG. 1A;
[0013] FIG. 3 is a cross-sectional view of a portion of the
sensor;
[0014] FIG. 4 is a plot showing two successive voltage ramps and
the bridge response corresponding to each ramp; the first response
corresponds to an analyte-loaded bridge and the second response
corresponds to an identical but unloaded bridge;
[0015] FIG. 5 shows a thermal response profile obtained by
subtracting the second response of FIG. 4 (from the unloaded
bridge) from the first response of FIG. 4 (from the analyte-loaded
bridge);
[0016] FIG. 6 shows thermal response profiles for three explosives
heated in three separate experiments at different adsorbed mass
levels;
[0017] FIG. 7 shows thermal response profiles for two
non-explosives heated in two separate experiments at different
adsorbed mass levels;
[0018] FIG. 8A shows thermal response profiles for the explosive
and non-explosive analytes shown in FIGS. 6 and 7;
[0019] FIG. 8B is a close-up view of the lower right hand corner of
the plot of FIG. 8A;
[0020] FIGS. 9A-9D are atomic force microscope (AFM) images of
adsorbed analytes, including TNT (FIG. 9A), PETN (FIG. 9B), RDX
(FIG. 9C), and NH.sub.4Cl (FIG. 9D), on silicon oxide surfaces;
[0021] FIG. 10 shows the thermal response profiles for rapid pulsed
heating of TNT at different adsorbed mass levels; and
[0022] FIG. 11 is a comparison of the temperature decay rates of
several analytes in comparison with the characteristic decay, which
is the expected temperature return rate for a microbridge with no
adsorbed analyte.
DETAILED DESCRIPTION
[0023] A novel microbridge sensor that generates a unique thermal
response from an adsorbed chemical analyte to achieve chemical
selectivity without sacrificing sensitivity or reversibility is
described. Due to its micro- and sub-microscale dimensions, the
microbridge sensor may be heated to more than 500 degrees in 50
milliseconds by passing current through one or more heating tracks
embedded in its surface. A measuring track changes resistance with
temperature to provide a thermal response signal. The thermal
response spectrum or profile over the duration of the heating
provides a signature of the adsorbed analyte. The microbridge
sensor is capable of differentiating explosive from non-explosive
materials, and is further capable of differentiating individual
explosive molecules such as trinitrotoluene (TNT), pentaerythritol
tetranitrate (PETN), and cyclotrimethylenetrinitramine (RDX). The
microbridge sensor may also be applied to the detection of other
chemical analytes.
[0024] Referring to FIG. 1A, the sensor 100 includes a microscale
body 105 having a first end 105a and a second end 105b and a
surface 110 between the ends 105a, 105b. Having a bridge
configuration, the microscale body 105 is restrained at the first
end 105a and the second end 105b and is substantially isolated from
its surroundings between the two ends 105a, 105b.
[0025] The microscale body (or microbridge) 105 includes at least
one conductive heating track 115a for heating molecules adsorbed
onto the surface 110 and a conductive response track for producing
a thermal response signal from the adsorbed molecules. The
conductive heating track 115a preferably extends from the first end
105a to the second end 105b of the microbridge 105 and is
electrically connected to a voltage source 125a. The conductive
response track 120 is electrically isolated from the heating track
115a and is electrically connected with a signal recorder 135.
Preferably, the conductive response track 120 extends from the
first end 105a to the second end 105b of the microbridge 105.
According to the embodiment shown in FIG. 1A, the microscale body
105 includes two conductive heating tracks 115a, 115b disposed on
either side of the conductive response track 120. When the
microscale body 105 and any adsorbed analyte molecules are heated
by passing electrical current through the conductive heating tracks
115a, 115b, the resistance of the conductive response track 120
varies sensitively as a function of temperature. Each heating track
115a, 115b is joined to electrical contacts 130a, 130b connected to
the voltage source 125a, 125b and ground. The response track 120 is
joined to electrical contacts 140, which are connected to a signal
recorder 135 and ground.
[0026] The microbridge sensor 100 is shown as part of a Wheatstone
bridge circuit 145 in FIG. 1B. The Wheatstone bridge circuit 145 is
employed to measure small changes in the resistance of the
conductive response track 120. The output of the Wheatstone bridge
circuit 145 is transmitted to the signal recorder 135. The circuit
145 includes two fixed resistors R.sub.1 and R.sub.2 and two
variable resistors, R.sub.3 and R.sub.4. The fixed resistors
R.sub.1 and R.sub.2 are identical resistors with constant
resistance. Variable resistor R.sub.3 is used for sensing and its
resistance varies during the measurement. Variable resistor R.sub.4
is used to balance R.sub.3 before the measurement. The response
track 120 of the analyte-loaded microscale body 105 serves as
variable resistor R.sub.3 and the response track of an unloaded but
otherwise identical microbridge serves as variable resistor
R.sub.4. The configuration allows a differential measurement of the
resistance of the sensor to be made by comparing the resistance of
the analyte-loaded microscale body 105 to the resistance of the
unloaded microbridge, which is not exposed to the analyte
vapor.
[0027] If the Wheatstone bridge circuit 145 is balanced, namely if
R.sub.1=R.sub.2 and R.sub.3=R.sub.4, then the output of the
circuit, namely V.sub.1-V.sub.2, is zero. During the measurement,
the resistance of R.sub.3 varies, and thus V.sub.1 varies. Since
R.sub.2 and R.sub.4 are fixed in the measurement, V.sub.2 does not
vary, and thus the output of the bridge circuit, namely
V.sub.1-V.sub.2, is related only to the change of R.sub.3.
Generally, an amplifier is connected with the bridge circuit to
increase the gain. By using the Wheatstone bridge circuit 145, the
resistance change of R.sub.3 can be determined from the voltage
output V.sub.O.
[0028] It is possible to heat the microscale body to hundreds of
degrees centigrade in milliseconds due to the body's extremely low
thermal mass. Rapid and controllable heating rates, dT/dt, where T
is the temperature and t is the time, of up to 10.sup.8.degree.
C./s can be achieved with the sensor configurations described and
depicted herein. The heating rate may be in the range of from about
10.sup.3.degree. C./s to 10.sup.5.degree. C./s, and a heating rate
of about 10.sup.4.degree. C./s has been found to be particularly
advantageous for obtaining detailed thermal response profiles.
[0029] To achieve the desired heating rates, an appropriate voltage
is applied to the heating track over a short time period.
Generally, given the heat capacity and the track resistance of
embodiments of the microbridge sensor described and depicted
herein, an increasing voltage of up to 10 V is appropriate. For
example, a voltage that increases to a value in the range of from 5
V to 8 V may be applied to the heating track(s). The voltage may be
ramped up or pulsed over a time duration of at least about 10 ms
and typically no more than about 100 ms. The time duration of the
heating may be, for example, from about 30 ms to about 70 ms. It
may be advantageous to apply a linearly increasing voltage to the
heating track for a time duration in this range. Alternatively, the
voltage may be applied as a step function (pulsed) for a shorter
time duration (e.g., about 10 ms). The voltage may also be applied
at different rates (dV.sub.x/dt.sub.x) over the time duration of
the heating. For example, a voltage may initially be applied to the
heating track at a first rate dV.sub.1/dt.sub.1 and then applied to
the track at a second rate dV.sub.2/dt.sub.2 for a remainder of the
time duration of the heating.
[0030] The tunable heating rate dT/dt makes it possible to achieve
chemical speciation for sub-nanogram quantities of material without
relying on receptors or separation methods. Unlike the deflagration
of adsorbed explosives on cantilevers, which fails to achieve
speciation, the controlled heating of chemical analytes on
microbridge sensors provides very high chemical selectivity.
[0031] The low thermal mass of the microbridge, which facilitates
the rapid heating of the chemical analyte, can be attributed to its
extremely small dimensions (e.g., submicroscale thickness).
Presently fabricated sensors include a thin rectangular microscale
body, and thickness is a key parameter in determining the
sensitivity of the device. The detection sensitivity depends on how
effectively heat released from or absorbed by the deposited analyte
can raise or lower the temperature of the microscale body. A
thinner device has less heat capacity, and thus it can be heated or
cooled with a smaller amount of adsorbed material. On the other
hand, the thickness of the sensor is preferably large enough to
impart structural stability to the device. For example, the
microscale body is preferably at least about 10 nm (0.01 micron) in
thickness, and may advantageously be at least about 50 nm in
thickness. It is also preferred that the thickness of the
microscale body is no more than about 700 nm. It may be
particularly advantageous for the thickness of the microscale body
to be about 500 nm or less, about 300 nm or less, or about 100 nm
or less. For example, the thickness of the microscale body of the
sensor may range from about 10 nm to about 700 nm, from about 50 nm
to about 500 nm, or from about 100 nm to about 300 nm.
[0032] It is also desirable that the area of the surface of the
microbridge onto which the analyte is adsorbed is optimized. If the
surface area (in particular, the length) of the sensor is too
small, the heating of the bridge may be diminished due to end
effects, with the restrained ends of the sensor acting as
relatively massive heat sinks. In addition, the amount of analyte
that may be adsorbed onto the sensor decreases as the area of the
adsorbing surface is reduced. For example, if the length of the
sensor is halved compared to the original length, the surface area
available to adsorb the chemical analyte is correspondingly
reduced. Hence, despite the reduction in thermal mass gained by
reducing the length, the sensitivity of the device may not increase
since the amplitude of the response signal is related to the amount
of the chemical analyte adsorbed onto the microscale body. On the
other hand, a very large surface area that can adsorb a large
amount of analyte may correspond to a microscale body that is too
massive to heat at the desired rapid rates.
[0033] According to one embodiment, the surface of the device spans
an area in the range of from about 0.01 mm.sup.2 to about 0.1
mm.sup.2. It may be particularly advantageous for the surface area
of the device to be in the range of from about 0.03 mm.sup.2 to
about 0.07 mm.sup.2. For example, the surface area may be about
0.05 mm.sup.2.
[0034] Microbridges having a length of 300 microns, 400 microns, or
500 microns and a width of about 100 microns have been shown to
work effectively. The length must be sufficient to effectively heat
the bridge, as discussed above, and the width must be sufficient to
allow the conductive tracks to be electrically insulated from each
other by one or more regions of insulating material. Accordingly,
the length of the microscale body from the first end to the second
end may be in the range of from about 300 microns to about 500
microns, and the width of the microscale body may be in the range
of from about 50 microns to about 150 microns, although other
lengths and widths are also possible. A width in the range of from
about 50 microns to about 100 microns may be advantageous, for
example.
[0035] The sensitivity of the device may be enhanced by
micromachining bridges having a decreased thermal mass in
combination with a large surface area. It is contemplated that the
microscale body may have configurations other than that shown in
FIGS. 1 and 2 that provide a small thermal mass in conjunction with
an optimized surface area. For example, the microscale body may
have a cylindrical surface for adsorbing a chemical analyte. Other
non-flat or non-smooth surface geometries (e.g., curved, roughened,
corrugated, etc.) that allow the surface area to be maximized for a
given microbody configuration are also possible.
[0036] To facilitate uniform heating of the surface of the sensor,
the heating track(s) preferably cover substantially all of the
surface except for the area occupied by the response track and the
insulating regions. The heating track may have a width of from
about 30 microns to about 40 microns and extend at least from the
first end of the microscale body to the second end. The measuring
track may have a width of from about 15 microns to about 25 microns
and may also extend across the entire length of the microscale
body. It is also contemplated that the heating and response tracks
may have a curved, bent, or other non-straight or non-flat
configuration.
[0037] The conductive heating and response tracks may be embedded
into, deposited on, or otherwise included at or near the surface of
the microscale body. For example, dopant atoms (e.g., boron) may be
implanted into a silicon surface to form doped silicon of a
suitable conductivity. The conductive tracks may then be defined by
reactive ion etching, followed by deposition of silicon-rich
nitride for insulation and then a polysilicon layer, as described
in greater detail in the example below. FIG. 2 is a scanning
electron microscope (SEM) image of an exemplary microbridge sensor
fabricated from a silicon-on-insulator (SOI) wafer. The microbridge
sensor is 450 microns in length, 100 microns in width, and 550 nm
in thickness. Three electrically conducting tracks formed of doped
silicon are buried in the wafer and separated from each other by
thin layers of insulating silicon-rich nitride. The two outer
tracks are the conductive heating tracks 115a, 115b and the inner
track is the conductive response track 120, as shown schematically
in FIG. 3.
[0038] A silicon body including conductive tracks formed of
boron-doped silicon is described due to the ease of and/or
availability of technology for fabricating the microbridge sensor
from such materials. The sensor is not limited, however, to a body
formed of silicon or to conductive tracks formed of doped silicon.
The body may be formed of another material, such as a dielectric
material, for example, and the conductive tracks could be formed of
a metal or alloy deposited on or embedded into the dielectric
material. There may be other suitable ways of forming (and
materials from which to form) the body and conductive tracks of a
microbridge sensor having a small thermal mass.
[0039] In an exemplary fabrication process, the microbridge sensor
is constructed on a SOI wafer with a 400 nm buried oxide and a 340
nm device layer. In this example, the device layer is thinned to
200 nm by dry thermal oxidation, followed by boron doping to an
active level of 710.sup.19 cm.sup.-3 using ion implantation. The
conductive tracks are defined by reactive ion etching (RIE). A 250
nm layer of silicon-rich nitride is deposited by low pressure
chemical vapor deposition (LPCVD) to provide electrical insulation
between the tracks and to support the structure. A 90 nm layer of
poly silicon is also deposited by LPCVD. The bridge and contact
holes are defined using RIE and nitride etching (using phosphoric
acid at 180.degree. C.). The nitride on the back side of the wafer
is patterned by RIE and the structures are released by a KOH etch
at 80.degree. C. During the release, the front side of the wafer is
mechanically protected. Finally, a metal layer of Ti/Au is
deposited and wires are defined by an etch sequence (KI, I.sub.2
and HF). The microbridge fabricated in this exemplary process is
500 microns long and 100 microns wide. The two heating tracks are
35 microns in width and the measuring track in the center of the
bridge is 10 microns wide. The heating tracks have a resistance of
1.77 k.OMEGA. and the measuring track has a resistance of 4.96
k.OMEGA.. The doping level affects the resistance of the conductive
tracks. A high doping level is selected to permit the bridge to be
heated to a higher temperature before the intrinsic doping
concentration interferes with the response signal.
[0040] The microbridge device may be kept in the open air directly
in an explosive plume for evaluation or usage of the sensor. For
evaluation purposes, the explosive plume may be produced by a
custom-made vapor generator with a heated outlet, any example of
which is described by Pinnaduwage et al. in Langmuir 2004, 20,
2690-2694. Explosive vapors created at elevated temperatures
condense on the microbridge, which is maintained at room
temperature. A voltage ramp is then used to increase the
temperature of the microscale body at a desired high heating rate
dT/dt. During heating, the adsorbed molecules melt, evaporate,
decompose and/or desorb as a function of temperature, changing the
resistance of the response track. For a dT/dt value of, for
example, about 1.1.times.10.sup.4.degree. C./s, the thermal
response profile has a unique shape that depends on the adsorbed
explosive.
[0041] The rate of heating depends on the mass of the bridge.
Therefore, a bridge with added mass (adsorbed molecules) will have
a different heating rate than an otherwise identical but unloaded
bridge. Since all the adsorbed mass leaves (desorbs) from the
loaded bridge by the end of the heating cycle, the thermal response
obtained from a second heating step applied to the same bridge can
serve as a baseline signal representing the bridge with no adsorbed
mass.
[0042] FIG. 4 shows a linear voltage ramp of from 0 to 6.4 volts
applied over a period of 50 milliseconds to the analyte-loaded
bridge and then to the unloaded bridge. In this example, the
analyte is PETN. A first voltage ramp 405a is applied to the
PETN-loaded bridge to heat and ultimately desorb the PETN from the
surface, producing a first thermal response profile 410, measured
in terms of voltage. After the first heating cycle, the PETN is
believed to be completely desorbed from the surface. A second
thermal response profile 415 can then be obtained for the unloaded
bridge during a second voltage ramp 405b.
[0043] Referring to FIG. 5, a thermal response profile 505 for the
adsorbed chemical analyte in terms of voltage is created by
subtracting the signal 410 of the PETN-loaded bridge from the
signal 415 of the unloaded bridge.
[0044] The differential signal measured (see FIG. 5) as a function
of heating time is proportional to the difference in resistance
between the analyte-loaded bridge and the unloaded bridge, which is
in turn proportional to the difference in their temperatures. A
positive differential signal indicates that the temperature of the
unloaded bridge is higher than that of the analyte-loaded bridge.
The measurement is also proportional to the rate of change of
thermal mass of the analyte-loaded bridge due to thermal desorption
of the adsorbed molecules (dM/dt). The mass loss rate (dM/dt) is
related to the product of dM/dT and the heating rate dT/dt. Since
the observed signal is with respect to the same bridge, any
mechanical buckling of the bridge under thermal stress does not
play a role in the signal production. The thermal response profiles
obtained from the differential measurements indicate how far the
temperature of the loaded bridge lags behind that of the unloaded
bridge. Explosives and other analytes may be distinguished using
the shapes of their thermal response profiles.
[0045] Referring to FIG. 6, the three separate line curves plotted
for each explosive correspond to three separate sensing experiments
with varying amounts of explosive. The sensor output is plotted as
a function of time (x-axis) due to the application of a linear
heating voltage ramp (5.4 V in 50 milliseconds) with an average
dT/dt of 10.sup.4.degree. C./s for each explosive. The explosives
employed to obtain the curves 610, 620, 630 were TNT, PETN, and
RDX, respectively. The three different curves (6X0a, 6X0b, and
6X0c; X=1, 2, or 3) shown for each explosive represent,
respectively, 0.6, 1.2, and 2.4 nanograms of adsorbed explosive on
the sensor. It can be seen that the amplitude of the response
varies as a function of the mass of adsorbed explosive, but the
shape of the response remains constant for a given explosive.
[0046] It is possible to obtain a thermal response from as little
as a few picograms of adsorbate. A more sensitive device can obtain
a signal from a smaller amount of adsorbate, but to obtain a
detailed response profile during a voltage ramp, a larger amount of
adsorbate may be advantageous. For example, 1 picogram to 6
picograms of adsorbate may be sufficient when a short (e.g., 10 ms)
voltage pulse is applied to the microbridge, but 100 picograms to
600 picograms is preferred with a longer duration voltage ramp
(e.g., over 50 ms). Generally, the amount of adsorbed chemical
analyte is in the range of from about 1 picogram to about 1000
picograms (1 nanogram).
[0047] FIG. 7 shows the thermal response of a bridge with adsorbed
non-explosive molecules obtained with the same dT/dt as above. The
non-explosive analytes employed to obtain the curves 710 and 720
were NH.sub.4Cl and Na.sub.2B.sub.4O.sub.7, respectively. The
response curves of the non-explosives appear as simple Gaussian
peaks characteristic of evaporation events, where the peak
positions vary with the specific material as well as the amount of
material adsorbed. The two separate line curves (7X0a and 7X0b; X=1
or 2) plotted for each analyte correspond to two separate
experiments with increasing mass of analyte, respectively. Unlike
the thermal response profiles for explosives, the non-explosive
profiles are featureless except for the Gaussian curvature.
[0048] In FIG. 8A, the response curves of FIGS. 6 and 7
corresponding to the largest adsorbed mass for each analyte (TNT,
PETN, RDX, NH.sub.4Cl and Na.sub.2B.sub.4O.sub.7) are shown
together for comparison. The response profiles for each explosive
show distinct shapes due to a combination of melting, evaporation,
and decomposition, while the response profiles for the
non-explosive analytes have simple Gaussian shapes. Explosive
responses tend to overshoot at the end of the response curve and
dip to negative values, as shown in FIG. 8B, which is an enlarged
view of the lower right hand corner of FIG. 8A. A signal below the
dashed line indicates that the temperature of the bridge with
adsorbed explosive is higher than the temperature of the bridge
without the adsorbate present. This overshooting suggests that an
exothermic decomposition of the adsorbed analyte has occurred. In
contrast, the response profiles corresponding to the non-explosive
analytes do not fall below the dashed line, which indicates that
the temperature of the bridge with adsorbed non-explosive is always
lower than the temperature of the bridge without the adsorbate
present.
[0049] FIGS. 9A-9D are atomic force microscopy (AFM) images of
adsorbed analyte on a silicon oxide surface. The AFM images show
nucleation islands of adsorbed TNT (FIG. 9A), PETN (FIG. 9B), and
RDX (FIG. 9C) on a silicon oxide surface of 10 square microns in
area. The size of these islands increases with increasing exposure
to the analyte.
[0050] Referring again to FIG. 8A, the response profile 610 for TNT
shows a single peak with a slow rise and fast fall. The slow rise
is consistent with melting and evaporation of nucleated TNT islands
on the surface. Small islands evaporate faster than larger ones.
AFM images of nucleated islands of TNT as a function of time at
room temperature show residues left behind, probably crystalline
TNT. It is believed that fast evaporation of islands leaves behind
crystallites that exothermically decompose at higher temperature.
The fast fall in the response curve is related to the rate at which
the bridge's temperature is increasing, which may be attributed to
a combination of exothermic decomposition and a decrease in thermal
mass as material leaves the surface.
[0051] The shapes of the thermal response profiles 620, 630 of RDX
and PETN shown in FIG. 8A include a slow rise followed by a rapid
rise to a distinct double peak, and then a fast fall. The initial
slow rise can be attributed to the evaporation of nucleated islands
of explosives, as discussed above. However, for the PETN and RDX
response curves 620, 630, a rapid rise to a peak follows the slow
rise at the melting point for each explosive. This peak may be due
to the absorption of thermal energy needed to melt the PETN and
RDX. The fall to the second peak followed by the fast fall is
believe to be related to the exothermic decomposition of the
explosives coupled with a decrease in the thermal mass, as is the
case for TNT.
[0052] The temperatures at which the response peaks or features
occur are different for different explosives, and the overshooting
in the response profiles as shown in FIG. 8B is observed only for
explosives. A combination of characteristic shapes and overshooting
due to an exothermic process can serve as a clear indicator of a
particular explosive analyte. In contrast, non-explosive analytes
have thermal profiles of relatively simple shapes that indicate a
pure evaporation from a uniform layer of adsorbed material.
Referring to FIG. 9D, the AFM image of NH.sub.4Cl adsorbed on a
silicon oxide surface of 10 square microns in area shows uniform
coverage with a finite surface roughness, in contrast to the
explosives imaged in FIGS. 9A-9C. The height scale of the AFM
figures is 10 nm from dark to light. Mass loss due to evaporation
from a uniform surface coverage is expected to be a sigmoid and the
rate of evaporation is expected to follow a Gaussian profile, as
observed in FIG. 7.
[0053] To illustrate high specificity in detection, sensing
experiments have been conducted with interferents such as
non-explosives, as discussed above, volatile organic compounds
(VOCs), and water vapor. Experiments conducted with VOCs did not
produce any discernable signals. One possible explanation is that
the amount of mass adsorbed is less than the detection threshold.
Since the VOCs have smaller sticking coefficients than the
explosives, the mass loading of adsorbed VOCs on the microbridge
sensor is expected to be lower.
[0054] To determine the effect of relative humidity (water vapor)
on sensor performance, experiments were carried out at different
humidity levels. Since the explosive vapor concentration may be
millions of times smaller than the water vapor concentration in
air, relative humidity may be a significant interferent to chemical
vapor detection. Even at 80% relative humidity, which was the
highest humidity level tested, adsorbed water molecules did not
produce a thermal response signal. The results indicate that the
microbridge sensor can function even in the presence of high
humidity.
[0055] The characteristic shapes of the nanothermal profiles may be
significantly influenced by the value of dT/dt. At higher dT/dt
values, the observed peaks or features merge into a single peak,
losing their speciation characteristics. FIG. 10 shows the
nanothermal profiles obtained from a microbridge loaded with TNT
when rapidly heated to 8 volts by a 20 millisecond square wave
pulse. The responses for adsorbed TNT do not show any features as
observed with ramped heating. Similar unresolved response profiles
were obtained for fast heating of other explosives. FIG. 10 shows
that under rapid heating, the peak temperature does not shift with
the amount of explosive adsorbed. This is consistent with the
evaporation rate of individual islands. The nonspecific response
observed for explosives under rapid heating has a limit of
detection (LOD) of 6 picograms, which is sufficient to detect even
RDX at room temperature within 10 seconds of sampling time without
the use of a preconcentrator. It is possible to combine pulsed and
ramped heating approaches in such a way that ramped heating is
initiated for speciation only if pulsed heating shows a response. A
combined approach may yield a selective sensor for the detection
and identification of explosives that is rapid, sensitive, and
completely reversible. Additionally, by varying the voltage versus
time function, different regimes can be highlighted. A slow
increase in voltage emphasizes the evaporative regime. A rapidly
increasing voltage generally leads to deflagration prior to
complete sublimation. A pulsed current gives an integrated response
that is indicative of the analyte formation enthalpy. It is
contemplated that the microbridge sensor could be heated with a
gradient as high as a few million degrees per second. An
intermediate time scale (e.g., based on a voltage ramp) displays
the whole range of phenomena.
[0056] The mass of adsorbed explosive vapor can be calculated using
a resonating microcantilever beam placed in the same plane as the
bridge sensor. For a cantilever free at one end, the frequency
decreases as the adsorbed mass increases according to the relation
.DELTA.m/m=-2.DELTA.f/f, where m and f represent the mass and
resonant frequency of the cantilever, respectively. For the present
bridge structure, the relationship is further complicated by the
fact that both ends of the bridge are attached to the supporting
substrate, and thus the surface stress resulting from the
adsorption of mass alters the spring constant of the device. This
makes it difficult to separate changes in resonant frequency due to
mass loading (decreasing f) from changes due to surface stress
(increasing f). The resonance frequency of the bridge structure can
be measured by using an optical beam deflection method, which shows
frequency increasing as a function of vapor adsorption due to
adsorption-induced surface stress effects. Therefore, the mass of
explosive vapor adsorbed on the structure may be estimated using a
resonating reference cantilever device after exposure to the same
duration of analyte vapor as used for the bridge device. For
calibration purposes, the mass adsorption can be assumed to be
uniform across the cantilever and the effective mass/area may be
calculated. It also can be assumed that the mass/area is the same
for the bridge sensor, and therefore the mass adsorbed on the
bridge can be calculated as the product of the cantilever adsorbed
mass/area with the area of the bridge surface.
[0057] From the resonance frequency measurements, the calculated
limit of detection (LOD) for the fabricated microbridge device
shown in FIG. 2 is estimated as approximately 0.6.times.10.sup.-9 g
(600 picograms) of adsorbed mass, which is slightly less than 1% of
the inertial mass of the bridge and corresponds to about 10.sup.11
molecules. Since the vapor pressures of explosives are typically
extremely small (e.g., the vapor pressure of PETN is around 5
parts-per-trillion (volume) at room temperature), a preconcentrator
may be advantageous for collecting explosive molecules on the
bridge surface. Preconcentrators sample large volumes of air to
collect explosive molecules and/or other particulates. The
collected molecules are then abruptly desorbed, at which time the
explosive concentrations may be at ppb levels or higher for
detection. High efficiency preconcentrators may be able to collect
and concentrate low vapor pressure explosives. Exemplary
preconcentrators are discussed in "Miniaturized explosives
preconcentrators for use in man-portable explosives detection
systems," by D. W. Hannum et al., in the Proceedings of the IEEE
34th Annual 2000 International Carnahan Conference on Security
Technology, pp. 222-227, which is hereby incorporated by reference
in its entirety. Explosives such as TNT have a relatively high
vapor pressure in the parts-per-billion range, and thus may benefit
from a mild preconcentration.
[0058] The ramped heating method can provide clear thermodynamic
signatures which vary significantly with analyte and are therefore
useful for analyte identification, even without further
interpretation of the thermodynamic information contained in the
response curves. However, a virtue of the method with regard to
signaling the presence of energetic molecules is that it is
sensitive to molecular energies. An understanding of the response
in terms of thermodynamics may lead to general identification of
energetic materials, even those whose signals have not been
previously catalogued. Furthermore, a thermodynamic understanding
may guide the design of the bridge sensors to optimize selectivity.
Without wishing to be bound by theory, the following thermodynamic
analysis of the microbridge sensor is set forth.
[0059] The unloaded bridge can be modeled by a total heat capacity,
C (J/K), an electrical conductivity G (J/sN.sup.2) and an average
"cooling/heating" coefficient, k (J/K/s). The unloaded bridge
temperature is well described by the simple linear differential
equation:
C T _ t = G ( V ( t ) ) 2 - k ( T _ - T a ) ( 1 ) ##EQU00001##
[0060] where T.sub.a is the ambient temperature (324K) and T is the
average temperature of the bridge. When the voltage is linear in
time, i.e. V(t)=rt, the exact solution is:
T _ - T a = a [ ( t .tau. ) 2 - 2 ( t .tau. ) + 2 ( 1 - exp ( - ( t
.tau. ) ) ) ] where .tau. = C / k and a = r 2 G k .tau. 2 . ( 2 )
##EQU00002##
[0061] The linearity of Eq. (1) and the linearity of the thermistor
justify the assumption that the thermistor temperature, T.sub.T,
and the analyte temperature, T.sub.A, are proportional to the
average bridge temperature:
T T = a T a T _ and T A = a A a T _ . ##EQU00003##
Hence Eq. (2) also applies to the thermistor temperature, T.sub.T,
or the average temperature of the analyte, T.sub.A. The function
T(t) is simply multiplied by
a T a or a A a ##EQU00004##
to obtain T.sub.T and T.sub.A respectively. Fitting the two
constants, a.sub.T and .tau. to the T.sub.T(t) of the unloaded
bridge gives a.sub.T=1.7K and .tau.=2.6 ms. Note that the time
constant, .tau., of the bridge is a robust quantity that describes
the exponential decay of temperature excursions ( T, T.sub.T, or
T.sub.A) back to the ambient temperature.
[0062] Next, consider adsorbed analyte droplets on the bridge
within a temperature range where the only transformation is to the
vapor (evaporation or sublimation). The term vaporization is used
to indicate either sublimation or evaporation. Eq. (1) is modified
by the cooling due to the loss of analyte mass to vapor:
C T _ t = G ( V ( t ) ) 2 - k ( T _ - T a ) + V m t ( 3 )
##EQU00005##
[0063] where, .epsilon..sub.V is the enthalpy of vaporization.
Taking the difference between Eqs. (1) and (3) gives:
C T _ m 0 - T _ 0 t = - k ( T _ m 0 - T _ 0 ) + V m t C .DELTA. T t
= - k ( m ) .DELTA. T + V m t ( 4 ) ##EQU00006##
[0064] In Eq. (4) a cancellation of terms is achieved between the
loaded and unloaded bridge that is similar to the cancellation
occurring in the experiment. In Eq. (4), the heat flow that is
controlled by k now returns .DELTA..sub.T to zero. By comparing the
solution of Eq. (4) to the measurements, it is found that more
realistic values are obtained by assuming that the coating of
analyte on the surface reduces the coefficient k. We postulate that
the dependence of k on the load is:
k(m)=k(m=0)(1-.alpha.m.sup.2/3) (5)
[0065] where k(m=0) is the value determined from the unloaded
bridge. The process is assumed to be activated with an activation
energy .epsilon..sub.A and to have a vaporization rate proportional
to the radius of the droplet:
m t = - .gamma. T a exp ( - A k B T A ( t ) ) m 1 3 ( 6 )
##EQU00007##
[0066] The temperature controlling the activation is not .DELTA.T
but the temperature of the analyte, T.sub.0+.DELTA.T. The
sublimation (evaporation) enthalpies for TNT, RDX, and PETN are 498
J/g (402 J/g), 588 J/g (424 J/g), and 476 J/g (422 J/g)
respectively. These enthalpies set the prefactor of the vaporized
mass term in Eq. (4). The prefactor, .gamma..sub.T.sub.a, and
activation energy, .epsilon..sub.A, are not readily available but
can be selected to give reasonable agreement with the evolution of
the response curves below the melting points of RDX and PETN. For
TNT, there is no indication of melting, and thus it is assumed that
the TNT is liquid or glass through the entire measurement. The
response of TNT therefore illustrates the behavior of the bridge
subject to vaporization only. There is a transient stage lasting 4
ms during which evaporative cooling is either suppressed by surface
contamination or is counter balanced by weak exothermic reactions
related to annihilation of defects. This is followed by a period of
linearly increasing .DELTA.T as the explosive evaporates at the
rate appropriate to the ambient temperature. As .DELTA.T grows,
heat flow controlled by k contributes a quadratic term that reduces
.DELTA.T from linear growth. This is balanced by increasing
activation of the vaporization process, which is controlled by the
temperature of the bridge. The temperature initially grows slowly,
but then accelerates. Eventually, the accelerating rate of
evaporation causes .DELTA.T to increase exponentially. The rapidly
increasing vaporization results in a sudden depletion of the
analyte. The peak in .DELTA.T occurs when the cooling rate due to
the vaporization of a small amount of remaining analyte matches the
heat flowing in. Soon afterwards the analyte is essentially gone,
and, according to Eq. (4), the temperature should increase back to
T.sub.0 with time constant, .tau..
[0067] Actually, for TNT in the last stage, the observed behavior
deviates from Eq. (4); the temperature increases faster than
indicated by .tau.. This may be attributed to an exothermic
transformation turning on as k.sub.BT becomes comparable to the
activation energy for deflagration (621 J/g). The enthalpy of
deflagration (4560 J/g) is much higher than both the vaporization
and melting enthalpies; therefore, even a small amount of
deflagration can cause a return of .DELTA.T to zero at a faster
rate than the natural decay rate of the bridge. In fact, decay of
.DELTA.T faster or slower than indicated by the characteristic
temperature decay time, .tau., unambiguously indicates exothermic
or endothermic reactions (reverse if .DELTA.T is negative). This is
illustrated in FIG. 11, where responses are aligned in time and
normalized to facilitate comparison to the characteristic decay.
The characteristic decay (curve 1100) is the expected temperature
return rate for a bridge with no load. The non-explosive NH.sub.4Cl
(curve 1710) has a rate slower than the characteristic decay due to
thermal loading, while the explosives TNT (curve 1610), PETN (curve
1620), and RDX (curve 1630) all show return rates faster than the
characteristic decay due to added heat from exothermal
deflagration.
[0068] In RDX, PETN, and NH.sub.4Cl melting occurs. Melting results
in a cooling of the sample proportional to the enthalpy of fusion
that can be appended to Eq. (4)
C .DELTA. T t = - k .DELTA. T - V m t - F m liquid t ( 7 )
##EQU00008##
[0069] Melting is assumed to take place instantaneously as a region
of the analyte reaches the analyte melting point. Melting is
initiated when the hottest part of the load reaches the melting
point and continues until the coolest part of the load has reached
the melting point. This occurs over a fairly short period of time.
Therefore, the jump in .DELTA.T due to melting can be estimated by:
.DELTA.T.sub.Melting=.epsilon..sub.Fm/C.
[0070] Another feature of the data is a noticeable shift in the
response as the load is increased. For larger loads, additional
time (delay time) is required for the electric current to add the
amount of Joule heat needed to overcome the cooling effect of
vaporization and melting. The increase in heat capacity due to the
specific heat of the load is found to have a very small effect on
the delay. For explosives, the major contributor to the delay is
the vaporization cooling. This energy loss to vaporization cooling
is approximately equal to C.DELTA.T and the delay .DELTA.t.sub.v is
given by this energy divided by the rate at which energy is
accumulating in the bridge from Joule heating:
.DELTA. t V = C .DELTA. T ( CT 0 t ) - 1 = .DELTA. T ( T 0 t ) - 1
( 8 ) ##EQU00009##
[0071] The time delay due to melting can be estimated in a similar
way:
.DELTA. t Melting = V m liquid ( T 0 t ) - 1 ( 9 ) ##EQU00010##
[0072] For the explosives, the vaporization time delay is larger
than the melting delay. For the NH.sub.4Cl, the melting delay is
larger. Adjusting the response curves to take these delays into
effect gives alignment of response features corresponding to
different loadings.
[0073] The simple analysis presented here allows thermodynamic
quantities such as activation energies, enthalpies of fusion and
evaporation, melting points and deflagration to be identified in
the response curves. Further pursuit of this approach may yield
general guidelines for the detection of energetic molecules. The
range of analyte interactions affects the thermodynamic properties
as a function of analyte depth because the binding of the top layer
is different from that in deeper levels closer to the bridge
surface. Subtle effects of this type may reveal themselves with
more detailed modeling.
[0074] Although the microbridge sensor is believed to be
particularly advantageous for the identification and speciation of
explosive materials, its usage is not limited to explosive
detection. The technique could be employed for the detection of
other chemical analytes, such as those with high vapor pressures.
For example, the sensor bridges could be cooled to below ambient
temperature to enable enhanced adsorption or condensation of high
vapor pressure analytes from the vapor phase. Carrying out the
temperature cycling in an inert atmosphere may enable the
investigation of the oxidative stability of analytes. The technique
may also have applications in the pharmaceutical and polymer
industries for investigating the temperature-dependent properties
of sub-nanogram quantities of polymers and for evaluating the
purity of drugs.
[0075] The demonstrated sensitivity and selectivity of this
approach offer new possibilities for a single sensor-based
technique that does not use a chemoselective layer for sensing.
This method may also provide a technique for investigating
thermally-induced properties of a wide range of materials far
beyond what is possible by conventional techniques. A key advantage
is that the detection process can be repeated continuously with the
sensor regenerating to a pristine surface after each thermal cycle
without resorting to chemical cleaning techniques. The microbridge
devices described herein may provide thermal response data from
sub-nanogram levels of adsorbates in a rapid, simple, and low-cost
manner.
[0076] Although the present invention has been described in
considerable detail with reference to certain embodiments thereof,
other embodiments are possible without departing from the present
invention. The spirit and scope of the appended claims should not
be limited, therefore, to the description of the preferred
embodiments contained herein. All embodiments that come within the
meaning of the claims, either literally or by equivalence, are
intended to be embraced therein. Furthermore, the advantages
described above are not necessarily the only advantages of the
invention, and it is not necessarily expected that all of the
described advantages will be achieved with every embodiment of the
invention.
* * * * *